A high-molecular-mass cell-surface protein from ... - CiteSeerX

Microbiology (2002), 148, 433–442

Printed in Great Britain

A high-molecular-mass cell-surface protein from Lactobacillus reuteri 1063 adheres to mucus components Stefan Roos and Hans Jonsson Author for correspondence : Hans Jonsson. Tel : j46 18 673382. Fax : j46 18 673392. e-mail : hans.jonsson!mikrob.slu.se

Department of Microbiology, Swedish University of Agricultural Sciences, Box 7025, S-750 07 Uppsala, Sweden

A gene from Lactobacillus reuteri 1063 encoding a cell-surface protein, designated Mub, that adheres to mucus components in vitro has been cloned and sequenced. The deduced amino acid sequence of Mub (358 kDa) shows the presence of 14 approximately 200 aa repeats and features typical for other cellsurface proteins of Gram-positive bacteria. Fusion proteins consisting of different repeats of Mub and the maltose-binding protein (MBP) were produced. These proteins adhered to pig mucus components, with molecular masses ranging from T01 to S2 MDa, to pig gastric mucin and to hen intestinal mucus. The binding of Mub to mucus components occurred in the pH range 3–74, with maximum binding at pH 4–5 and could be partly inhibited by the glycoprotein fetuin. Affinity-purified antibodies against recombinant Mub were used in immunofluorescence microscopy to demonstrate the presence of Mub on the cell surface of strain 1063. By using the antibodies in a Western blot analysis, Mub could also be detected in the growth medium. The results implicate Mub as a cell-surface protein that is involved in Lactobacillus interactions with mucin and in colonization of the digestive tract.

Keywords : adhesion, mucin

INTRODUCTION

Members of the genus Lactobacillus are often found in the gastrointestinal tract of birds and mammals (Finegold et al., 1983 ; Drasar & Barrow, 1985). For bacteria to be able to colonize this open flow environment, adhesion to the mucosa is considered to be a prerequisite (Berg, 1996 ; Brassart & Schiffrin, 1997). The epithelial cells of the intestine are covered by a protective layer of mucus, which is a complex mixture of glycoproteins and glycolipids with the large glycoprotein mucin being the main component. Bacteria colonizing the mucosa can be found both in the mucus layer and adhering to the epithelial cells (Savage, 1980). During recent decades, several reports describing adhesion of lactobacilli to epithelial cells and mucus have been published (Conway & Kjelleberg, 1989 ; Coconnier et al., 1992 ; Adlerberth et al., 1996 ; Rojas & Conway, 1996). Relatively recently, Kirjavainen et al. (1998) showed that most lactobacilli .................................................................................................................................................

Abbreviations : CnBP, collagen-binding protein ; MBP, maltose-binding protein ; NANB, non-A, non-B. The GenBank accession number for the sequence reported in this paper is AF120104. 0002-4917 # 2002 SGM

used in probiotic products adhered to human intestinal mucus. In most cases, the adhesion has been reported to be mediated by proteins, but saccharide moieties on the cell surface of lactobacilli have also been described to interact with components of the mucosa (Henriksson & Conway, 1992 ; Mukai & Arihara, 1994). Despite several reports of the phenomenon, the identities of the bacterial adhesins involved are unknown, except for a collagenbinding protein (CnBP) of Lactobacillus reuteri (Aleljung et al., 1994 ; Roos et al., 1996). The binding activity of this protein is blocked by porcine intestinal mucin and by a lectin with specificity for α--galactose (Aleljung, 1994), suggesting that CnBP also adheres to mucin via a lectin-like interaction. Lactobacillus reuteri frequently occurs in the gastrointestinal tract of various mammals, including man (Molin et al., 1993 ; Naito et al., 1995). Strains of this species have been described to possess several properties believed to be important for its capacity to colonize. These are adherence to epithelial cells (Wadstro$ m et al., 1987), binding of fibronectin (Lindgren et al., 1992), expression of the cell-surface CnBP (Aleljung et al., 1994), production of an autoaggregation-promoting protein (Roos et al., 1999) and production of the 433

S. ROOS and H. JONSSON

antimicrobial substance reuterin (Axelsson et al., 1989). Recently, Jonsson et al. (2001) reported that many strains of Lb. reuteri adhere to components in mucus. Interestingly, for some of the strains the adhesion to mucus was triggered by addition of mucin to the growth medium. In this report we have characterized the adhesion of Lb. reuteri strain 1063 to mucus material in vitro. The results presented show that a high-molecular-mass protein, which possesses the typical features of cellsurface proteins from Gram-positive bacteria, adheres to components in mucus. METHODS Bacterial strains and growth conditions. Lactobacillus reuteri 1063 (ATCC 53608) and 1068 were previously isolated from the small intestine of a pig (Wadstro$ m et al., 1987). Escherichia coli LE392 was used as a λ phage host strain and E. coli TG1 was used for subcloning (Sambrook et al., 1989). Lactobacilli were grown at 37 mC in MRS broth (Oxoid). When growing lactobacilli for preparation of extracellular proteins, the semidefined substrate LDM II was used (Kotarski & Savage, 1979). E. coli was grown at 37 mC in LB broth (Sambrook et al., 1989) on a rotary shaker, or on LB agar. Ampicillin (Amp) was used at 50 µg ml−" for selection. Proteins and reagents. Lb. reuteri 1063 was grown for 16 h in 500 ml LDM II broth and the cells were harvested by centrifugation at 10 000 g. Proteins in the spent culture medium were precipitated by adding (NH ) SO to 40 % % g. The saturation and pelleted by centrifugation at% #15 000 proteins were dissolved in 50 mM acetic acid (HAc), dialysed (cut-off 10 000 Da) against the same buffer and subsequently lyophilized. The protein material was finally dissolved in 1 ml PBS (g l−" : NaCl, 8n0 ; KCl, 0n2 ; Na HPO .2H O, 1n44 ; # mC.% To# remove KH PO , 0n2 ; pH 7n3) and stored at k20 # % loosely bound surface material, the pelleted bacteria were washed five times with distilled water (50 ml each time). Thereafter, the wash solution was lyophilized, the material was dissolved in PBS (1 ml) and stored at k20 mC. Antibodies were raised against material from the culture medium and from cell-surface water wash in a rabbit immunized with approximately 100 µg protein and given three booster doses at 2-week intervals. The animal was sacrificed 8 weeks after the first immunization. To raise the specific activity of the antiserum (P108) against the mucus adhesion factor, it was preadsorbed against cells of non-adhering Lb. reuteri 1068. The bacteria were grown in 200 ml MRS for 16 h and washed twice in PBS supplemented with 0n05 % Tween 20 (PBST) after which they were suspended in 20 ml PBST. One millilitre of antiserum was mixed with 1 ml bacterial suspension and incubated at room temperature for 2 h. After centrifugation, the adsorbed antiserum was passed through a 0n2 µm filter. The IgG fraction from the adsorbed antiserum was purified on a Protein A-Sepharose CL-4B column (Amersham Pharmacia Biotech), according to the manufacturer’s instructions. Mucus from pig or hen was prepared from the small intestine of freshly slaughtered animals. The intestine was rinsed with cold PBST after which the mucus was released by gently scraping the mucosa and washing with PBST. Particles were pelleted by centrifugation and the mucus was stored at k20 mC. Urea was added (final concentration 8 M) to some of the mucus material after which it was fractionated on a Superose 6 PC 3.2\30 column in a Smart System (Amersham Pharmacia Biotech) at a flow rate of 50 µl min−". Fractions

434

(50 µl) were collected and stored at k20 mC. The relative molecular masses of the eluted fractions were estimated by calibration of the column with a Gel Filtration HMW Calibration Kit (Amersham Pharmacia Biotech). Binding of bacteria to immobilized mucus. Mucus material, fractionated mucus material, pig gastric mucin (Sigma ; M1778) and BSA were dissolved and diluted in 50 mM Na CO buffer, pH 9n7, and immobilized in microtitre wells # $ by incubation of 150 µl solution overnight at 4 mC (Greiner) with slow rotation. The final concentrations or amounts used were an OD of 0n1 for the mucus material, 20i dilution for #)! mucus and 100 µg ml−" for mucin and BSA. the fractionated The wells were blocked with 0n2 ml PBS supplemented with 1 % Tween 20 for 1 h and then washed with PBST. Lb. reuteri 1063 was grown in MRS broth for 16 h at 37 mC, washed once in PBST and diluted to an OD of 0n5 in the same buffer. A '!! was treated with proteinase portion of the bacterial suspension K (100 µg ml−") for 1 h at 37 mC and PMSF was added to a final concentration of 1 mM. Another portion was incubated with P108 antiserum IgG (30 µg ml−"), αMub1 (30 µg ml−"), αMub2 (30 µg ml−") (αMub1 and αMub2 are described below) or IgG from preimmune serum (30 µg ml−") for 1 h at 37 mC. The bacteria from the different treatments and untreated bacteria were washed four times with PBST and diluted to an OD of 0n5 in the same buffer. Bacterial suspensions (0n15 ml) '!! were added to each well and incubated for 1 h at room temperature. The wells were washed with PBST and the degree of binding was examined with an inverted microscope. The buffer was poured off and, after the wells had dried, OD was measured %!$ were done in in an ELISA plate reader. All measurements triplicate. Construction and screening of a λ library from Lb. reuteri 1063. Lb. reuteri 1063 was grown in 100 ml MRS broth and

DNA was extracted according to Axelsson & Lindgren (1987). The DNA was partially digested with Sau3AI and ligated into λ EMBL3 BamHI arms. Packaging into phage particles was performed according to the manufacturer’s instructions (Promega). After infection of E. coli LE392, the resulting plaques were screened with the adsorbed IgG-fraction from P108, according to the procedure described previously (Roos et al., 1996). Subcloning and isolation of positive clones. DNA from the λ

clones 108 : 21 and 108 : 34 was isolated according to Sambrook et al. (1989) and cleaved in separate reactions with EcoRI, HindIII, PstI, SalI, ScaI (complete cleavages) or Sau3AI (partial cleavage). The material from the five first cleavages was pooled, treated with T4 DNA polymerase to generate blunt ends and then ligated into a SmaI-cleaved pUC18 vector. The Sau3AI-cleaved material was ligated into a BamHI-cleaved pUC18 vector. Ligation mixes were electrotransformed into E. coli TG1 cells and the resulting clones were selected on LB agar plates containing ampicillin, followed by screening with the IgG fraction of the P108 antiserum as described previously (Roos et al., 1996). Plasmid DNA from positive clones was purified with the Wizard Minipreps DNA purification system (Promega) and characterized by restriction enzyme analyses and sequencing. Deletion clones of 108.21 : 3 were constructed by using the EraseABase kit (Promega), according to instructions of the manufacturer, and subsequently used in the sequencing work. DNA sequencing and analysis of the DNA and deduced protein sequences. The λ clones and different plasmid

subclones were used to determine the nucleotide sequence of the gene. Sequencing was performed by using the ABI PRISM Dye Terminator Cycle Sequencing Ready Reaction Kit (Perkin

Mucus binding of Lactobacillus reuteri

.................................................................................................................................................................................................................................................................................................................

Fig. 1. Map of protein Mub, the mub gene, the fusion proteins MBP-Mub1 and MBP-Mub2, and the different clones used in the work. S, Signal sequence ; RI–VI, type 1 repeats ; R1–8, type 2 repeats ; W, cell-wall-spanning region ; M, membranespanning region ; P, promoter ; T, transcription terminator.

Elmer) and commercial standard and customized sequencing primers. The sequencing samples were analysed on an automatic sequencing machine (ABI 377 ; Perkin Elmer). The \  DNA and protein data-handling package was used for analysis of the DNA and deduced protein sequence. The SignalP server (http :\\www.cds.dtu.dk\services\SignalP) was used for prediction of signal peptides. NCBI’s search tool  (http :\\www.ncbi.nlm.nih.gov\blast\) was used for similarity searches. Alignments of protein sequences were done with the   program (Thompson et al., 1994) and phylogenetic analyses with programs of the  package (Felsenstein, 1993).

dissolved in 0n1 ml PBS and stored at k20 mC. Fifty microlitres of the sample was fractionated on a Superose 6 PC 3.2\30 column in a SMART chromatography system (Amersham Pharmacia Biotech) at a flow rate of 50 µl min−". Fractions (50 µl) were collected, analysed by SDS-PAGE and stored at k20 mC. The relative molecular masses of the eluted fractions were estimated by calibration of the column with the Gel Filtration HMW Calibration Kit (Amersham Pharmacia Biotech). The amino acid sequence of the purified protein Mub was determined with an ABI 476A protein sequencer (Perkin-Elmer).

Production of fusion proteins and affinity-purified antibodies. Plasmid DNA from 108.34 : 1 (encoding Mub2 repeats)

preparations from the spent growth medium and the cell surface of strain 1063, Mub purified from strain 1063 and the two affinity-purified fusion proteins were mixed 1 : 1 with sample buffer containing SDS and 2-mercaptoethanol and separated by SDS-PAGE. Electrophoresis was performed on PhastGel Gradient 8–25 % gels with the PhastSystem (Amersham Pharmacia Biotech), according to the manufacturer’s instructions. The proteins were blotted to a Hybond-C nitrocellulose membrane (Amersham Pharmacia Biotech) by diffusion at 65 mC for 45 min (PhastSystem Development Technique File No. 220). The membrane was blocked in PBST for 1 h at 37 mC followed by an incubation overnight at 4 mC with the P108 antiserum (diluted 1\1000) or affinity-purified antibodies αMub1 and αMub2 (both diluted 1\1000). After washing with PBST, the membrane was incubated with HRPconjugated goat anti-rabbit IgG (Bio-Rad), diluted 1\1000, at 37 mC for 1 h. After washing, the membranes were developed with 4-chloro-1-naphthol as substrate. The gels were stained with Coomassie blue after the blotting procedure.

was cleaved at SmaI and PstI sites in the multilinker of the vector and 108.34 : 3 (encoding Mub1 repeats) with the NaeI site in the gene and the PstI site in the multilinker (Fig. 1). The first cleavage resulted in a 2031 bp fragment and the second in a 1247 bp fragment that were purified and ligated into an XmnI\PstI-cleaved pMAL-c2 vector (encoding the maltosebinding protein, MBP ; New England Biolabs). The constructs were electrotransformed into TG1 cells and transformants were selected on LB plates with 50 µg ampicillin ml−". A number of clones were analysed by restriction enzyme digestion and DNA sequence determination. Production of fusion protein was verified by SDS-PAGE and Western blot analysis. Fusion proteins (MBP-Mub1 and MBP-Mub2) were produced and affinity-purified according to New England Biolabs’ instructions. The concentrations of the proteins were estimated by measuring A and the purity by SDS-PAGE #)!were coupled to CNBr-activated analyses. The fusion proteins Sepharose 4B (Amersham Pharmacia Biotech) and used for affinity purification of the P108 antiserum mentioned above. The purified antibodies (αMub1 and αMub2) were transferred to PBS buffer supplemented with 1 % BSA and 0n05 % NaN and stored at 4 mC. Antibodies against MBP (αMBP) were$ purchased from New England Biolabs. Purification of Mub from strain 1063. Lb. reuteri 1063 cells

were cultivated for 20 h at 37 mC in 3 l LDM II broth supplemented with 2 % glucose. Bacteria were pelleted and the supernatant proteins precipitated with (NH ) SO (40 % % #g, dissolved % saturation). The precipitate was pelleted at 15 000 in 10 mM HAc and dialysed twice against the same buffer, after which the solution was lyophilized. The dry material was

SDS-PAGE and Western blotting of Mub. The protein

Immunofluorescence detection of Mub on the bacterial surface. Strains 1063 and 1068 were grown overnight in MRS

broth and washed once in PBST. The bacteria were attached onto Poly Prep slides (Sigma) by drying 10 µl bacterial suspension (OD of 1n0) onto the slide. The slides were '!! and blocked with PBST containing 5 % washed with PBST goat serum for 1 h at room temperature with slow rotation in a humidity chamber. After washing with PBST, the slides were incubated with a mixture of the affinity-purified antibodies αMub1 and αMub2, both diluted 1\30 in PBST. The slides were washed and incubated with FITC-labelled goat antirabbit F(abh) fragments (Sigma), diluted 1\40 in PBST, for 1 h # chamber. After washing, the presence of Mub in a humidity 435


was visualized by fluorescence microscopy. Controls where the primary antibodies were excluded were done in parallel. Adhesion of Mub to immobilized mucus. Mucus material, mucus fractions, mucin or BSA were immobilized in microtitre wells as described above. ELISA was performed by incubation of 150 µl λ lysate (produced according to Sambrook et al., 1989), diluted to an A of 0n2, 150 µl purified Mub, diluted to an A of 4 i 10−%, or#)!150 µl fusion protein or MBP diluted to an A#)! of 10−$, per microtitre well for 1 h at room tem#)! Normally the proteins were diluted in PBST, but perature. 10 mM citrate buffers, pH 3–6, with 0n15 M NaCl or 10 mM phosphate buffers, pH 6–7n4, with 0n15 M NaCl were used in some experiments. The wells were washed and incubated with the αMub1, αMub2 or αMBP antibodies, diluted 1\2000 in PBST, for 1 h at room temperature. After washing, the wells were incubated with an anti-rabbit IgG peroxidase conjugate (Bio-Rad), diluted 1\5000 in PBST, for 1 h at room temperature, washed and developed with 3,3h,5,5h-tetramethylbenzidine as substrate, according to Bos et al. (1981). A was %&! measured in an ELISA plate reader. All assays were performed in triplicate. Inhibition experiments were done by addition of different components to the microtitre wells together with the fusion proteins MBP-Mub1 and MBP-Mub2. The different components were mucin, fetuin, asialofetuin, BSA (each 1 mg ml−") or maltose, mannose, glucose, fucose, raffinose, N-acetylgalactosamine, N-acetylglucosamine and sialic acid (each 10 mg ml−"). SDS-PAGE and Western blotting of mucus material. The mucus material was mixed with sample buffer, separated by SDS-PAGE and blotted as described above. After blocking in PBST for 1 h at 37 mC the membrane was incubated overnight at 4 mC with MBP-Mub2, MBP-Mub1 fusion proteins or MBP (diluted to an A of 5 i 10−$ in PBST) followed by incubation for 1 h#)!at 37 mC with the affinity-purified antibodies αMub2 or αMBP (diluted 1\1000 in PBST). The incubation with the secondary antibody and development was performed as above.

RESULTS Adhesion of Lb. reuteri 1063 to mucus

Lactobacillus reuteri strain 1063 adhered efficiently to both pig and hen small intestinal mucus immobilized in microtitre wells. The bacteria also adhered to pig gastric mucin and to fractions of pig mucus material with molecular masses ranging from 0n1 to 2 MDa with maximum adhesion to the 0n1–0n25 MDa fraction. Treatment of strain 1063 with Proteinase K eliminated adhesion to mucus, suggesting that the binding was mediated by proteinaceous structures. The IgG fraction from P108, an antiserum raised against extracellular and cell-surface proteins from strain 1063, inhibited adhesion to pig and hen mucus by 95 and 85 %, respectively (Table 1). IgG from preimmune serum had no effect on the binding of the bacteria (data not shown). Thus, it was concluded that the antibodies could recognize the bacterial protein(s) involved in binding to mucus material. Cloning and characterization of mub

Using the P108 antibodies, raised against secreted proteins and cell-surface proteins from strain 1063, a λ library generated from the parental strain was screened. 436

A number of clones that were reactive with the antiserum were identified. Examination of the recombinant proteins produced by these clones showed that they were of different classes, as indicated by a comparison of band patterns in a Western blot analysis. Most of the clones, among them the two clones λ 108 : 21 and λ 108 : 34, were expressing proteins of 0n2 MDa (data not shown). Functional studies with an ELISA revealed that both clones expressed proteins that adhered to mucus (Table 2). Subcloning of the λ clones into a plasmid vector allowed identification of clones 108.21 : 3, 108.34 : 1 and 108.34 : 3, reacting with the antiserum and expressing peptides of different sizes. Sequence determination revealed that both λ clones indeed harboured the same gene, here designated mub (mucus binding). The gene in λ 108 : 21 is truncated at the 5h end, but its putative 3h end is present (Fig. 1). Expression of recombinant protein was possible because the gene was fused in-frame to the 5h region of another unrelated gene, as a coincidental result of the cloning procedure. This was confirmed by PCR analysis of chromosomal DNA from strain 1063 using primers hybridizing to both sides of the fusion site in clone λ 108 : 21 and also with primers designed from λ 108 : 34. Clone λ 108 : 34 contains the start of mub, but lacks the 3h-terminal 456 nt of the gene. The plasmid clones derived from λ 108 : 34 expressed parts of the gene, fused to lacZ in the plasmid vector. The combined sequence revealed an initiation codon, ATG, followed by an ORF of 9807 nt (GenBank accession no. AF120104) encoding a polypeptide, Mub, containing 3269 aa and with a deduced molecular mass of 358 kDa. The initiation codon is preceded by a possible Shine–Dalgarno sequence and, further upstream, by putative transcription initiation signals. A tentative transcription terminator is located downstream of the stop codon. Mub is predicted to have a 49 aa N-terminal secretion signal peptide. Thus, the mature protein has a predicted molecular mass of 353 kDa. In the C-terminal part of Mub, a cell-wall anchoring motif (LPQTG), a putative membrane-spanning region and a cell-membrane anchor are located. Mub is highly repetitive and contains two types of related amino acid repeats. Six copies (RI–VI) of a type 1 repeat (Mub1), consisting of 183–197 aa each, and eight copies (R1–8) of a type 2 repeat (Mub2), all 184 aa in length, were identified. These are organized in an interesting manner, with the Mub2 repeats R1–8 inserted into the Mub1 repeat RV (Fig. 1). The six Mub1 repeats are rather diverse, with amino acid sequence identities ranging from 15 to 85 %. The Mub2 repeats, on the other hand, are well conserved (91 % sequence identity), with R1, R3 and R5 being identical, and R2, R4 and R6 being identical. The highest sequence identity between Mub1 and Mub2 repeats was 35 %. The best match in a database search using the  search engine was an unknown protein from Lactococcus lactis subsp. lactis encoded by the gene ywfG (GenBank accession no. AAK06278). The protein has previously not been assigned any features, but it is possible to predict an N-terminal signal sequence and a

Mucus binding of Lactobacillus reuteri Table 1. Adhesion of Lb. reuteri strain 1063 with different treatments to immobilized mucus material .....................................................................................................................................................................................................................................

Each column represents the mean of triplicate determinationsp. Adhesion to wells coated with :

Adhesion (OD403) of :

Pig mucus Hen mucus BSA Blocked wells Pig gastric mucin Pig mucus material ( 2 MDa) Pig mucus material (1–2 MDa) Pig mucus material (0n5–1 MDa) Pig mucus material (0n25–0n5 MDa) Pig mucus material (0n1–0n25 MDa) Pig mucus material ( 0n1 MDa)

Untreated bacteria

Proteinase K-treated bacteria

P108 IgG-treated bacteria

0n169p0n023 0n136p0n004 0n012p0n007 0n002p0n001 0n120p0n002 0n077p0n019 0n060p0n016 0n138p0n001 0n106p0n005 0n181p0n003 0n127p0n014

0n007p0n004 0n007p0n003 0n002p0n001 0n001p0n001

0n001p0n001 0n020p0n006 0n002p0n001 0n001p0n001

Table 2. Adhesion of Mub and different segments of the protein to mucus material and mucin .................................................................................................................................................................................................................................................................................................................

Each value represents the meanp of triplicate determinations in an ELISA. Adhesion to wells coated with :

Adhesion (A450) of : λ108.21

λ108.34

lysate

lysate

Negative λ-lysate

Mub from strain 1063

MBP-Mub1

MBP-Mub2

MBP

Pig mucus 0n627p0n010 1n536p0n069 0n183p0n013 1n018p0n011 1n080p0n077 1n378p0n086 0n206p0n006 Hen mucus 1n111p0n051 0n039p0n006 0n716p0n076 BSA 0n110p0n006 0n119p0n009 0n015p0n002 Blocked wells 0n171p0n002 0n175p0n010 0n007p0n004 0n020p0n003 0n122p0n020 0n118p0n011 0n016p0n002 Pig gastric mucin 0n970p0n017 1n880p0n036 0n125p0n015 Pig mucus material ( 2 MDa) 1n102p0n341 1n984p0n194 0n084p0n004 Pig mucus material (1–2 MDa) 0n892p0n224 2n422p0n163 0n120p0n002 Pig mucus material 1n306p0n121 2n698p0n110 0n332p0n006 (0n5–1 MDa) Pig mucus material 0n872p0n023 1n549p0n137 0n188p0n003 (0n25–0n5 MDa) Pig mucus material 0n649p0n098 1n420p0n134 0n149p0n003 (0n1–0n25 MDa) Pig mucus material 0n912p0n037 2n150p0n205 0n077p0n007 ( 0n1 MDa)

C-terminal cell-wall anchoring motif (LPXTG), putative membrane-spanning region and a cell-membrane anchor. The protein consists of 926 aa and has four repeats of approximately 175 aa each. In a phylogenetic analysis, the repeats of this protein were shown to be more related to the Mub1 repeats (37 % sequence identity) than the Mub2 repeats are. The Mub repeats also showed similarities to a non-A, non-B (NANB) hepatitis virus antigen (Reyes et al., 1990) and the human ocular component hr44 (Braun et al., 1995). Some less pro-

nounced similarities were also found with a family of high-molecular-mass adhesion proteins from Haemophilus influenzae (Barenkamp & St Geme, 1996). Production and purification of the MBP-Mub fusion protein

To produce and purify recombinant Mub, two constructs encoding fusions of the MBP with Mub1 and Mub2 repeats, respectively, were constructed in the 437


(a)

(b) 1

2

3

4

5

1

2

3

4

5

kDa .....................................................................................................

200·0 116·25 97·4 66·2 45·0

protein expression vector pMAL-c2. MBP-Mub1 contains approximately two repeats whereas MBP-Mub2 contains almost four repeats and the molecular masses of the proteins are 94n4 and 123n2 kDa, respectively (Fig. 1). The purified fusion proteins were immobilized on Sepharose beads and used for affinity purification of the P108 antiserum. The antibodies obtained, αMub1 and αMub2, were used in Western immunoblot analyses and shown to react specifically with MBP-Mub1 and MBPMub2, respectively (data not shown). Both fusion proteins were subjected to degradation as shown on immunoblots (Fig. 2, lanes 4 and 5). Purification and characterization of Mub from Lb. reuteri

Mub could be detected in both spent culture medium and cell-surface extract from strain 1063 by immunoblot analysis (Fig. 2, lanes 1 and 2). Only one obvious band of such size was visible in the preparations and hence it could be purified in one step by gel filtration (Fig. 2, lane 3). The protein was eluted from the Superose 6 column at 0n93 ml, suggestive of a mass of approximately 2 MDa. The reason for this exceedingly high apparent molecular mass is not known, but it is possible that the protein forms aggregates or is linked to other structures of the bacterial cell wall. Two smaller proteins that were co-eluted with Mub might also be components of these supposed aggregates or structures. The N terminus of the purified protein was sequenced. The result showed a heterogeneous population of Nterminal sequences. However, by also looking at the second and third choice in some of the positions, the sequence of the predicted mature protein, ATTESNASAK, could be identified. This indicates that the predicted cleavage site is correct, but also that the protein is easily degraded. By using affinity-purified antibodies against recombinant Mub in an SDS-PAGE analysis (Fig. 2), it was evident that the large protein purified from strain 1063 was recognized by the αMub1 and αMub2 antibodies and most probably represents the full-length protein encoded by mub. By using the antibodies in an immunofluorescence microscopy experiment, Mub could also be 438

Fig. 2. SDS-PAGE/Western immunoblot analysis. (a) Coomassie-stained gel. (b) Immunoblot with αMub1 and αMub2 antibodies. Lanes : 1, proteins extracted from the cell surface of Lb. reuteri strain 1063 ; 2, proteins secreted from strain 1063 ; 3, Mub purified from growth medium ; 4, purified MBP-Mub1 ; 5, purified MBP-Mub2. An arrow indicates the position of full length Mub.

(a)

(b)

(i)

(ii)

.................................................................................................................................................

Fig. 3. Immunofluorescence detection of Mub on the surface of Lb. reuteri 1063 (i) or 1068 (ii). (a) Light microscopy and (b) immunofluorescence microscopy. Bar, 3 µm.

detected on the cell surface of strain 1063, but not on strain 1068 (Fig. 3). Adhesion of protein Mub to mucus material

In ELISA, recombinant Mub proteins as well as Mub purified from strain 1063 were shown to adhere to mucus (Table 2). To see to which component(s) Mub binds, adhesion to pig gastric mucin and fractionated pig mucus was measured. Both the MBP-Mub1 and the MBP-Mub2 fusions adhered to pig gastric mucin and to all tested mucus fractions with molecular masses ranging from 0n1 to 2 MDa, with maximum adhesion to the 0n5–1n0 MDa fraction. pH was shown to be an important parameter for adhesion. Although the interaction is efficient at pH 7n4, it is stronger at lower pH values between 3 and 6. MBP-Mub1 was more affected than MBP-Mub2 by a low pH and showed a maximum adhesion at pH 5 with a value more than 3n4 times higher than at pH 7n4. MBP-Mub2 showed maximum adhesion at pH 4 with a value more than 1n7 times

Mucus binding of Lactobacillus reuteri (a)

Inhibition of adhesion of Mub to mucus material

(b) 1

2

1

2

kDa 97·4 66·2 45·0 31·0 21·5 14·4

.................................................................................................................................................

Fig. 4. Adhesion of MBP-Mub2 to mucus material from pig or hen intestine. (a) SDS-PAGE ; (b) Western blot analysis. Lanes : 1, pig intestinal mucus ; 2, hen intestinal mucus.

Table 3. Inhibition of adhesion to mucus after addition of different substances (expressed as percentage of adhesion in absence of inhibition) Addition of :

Adhesion ( %) to mucus by : MBP-Mub1

No addition Mucin Fetuin Asialofetuin BSA Maltose Glucose Fucose Raffinose N-Acetylgalactosamine Sialic acid N-Acetylglucosamine Mannose

100 49 38 33 99 133 121 91 108 106 168 111 119

MBP-Mub2 100 21 7n5 6n0 81 121 109 91 102 107 118 118 95

higher than at pH 7n4. Additionally, changing from phosphate to citrate buffer in general increased the adhesion by approximately 40 % (data not shown). Adhesion of the MBP-Mub2 (Fig. 4) and MBP-Mub1 (data not shown) fusion proteins to mucus material was also demonstrated in Western blot analyses. Both fusion proteins showed a similar adhesion pattern, while the fusion partner MBP itself did not adhere. The fusion proteins were shown to adhere to more than 10 different components of pig mucus with molecular masses between 15 and 45 kDa (Fig. 4b), but also to components with very high molecular mass that could not enter the gel. Interestingly, they seemed to adhere only to a highmolecular-mass component in the hen mucus.

To further characterize the interaction between Mub and mucus components, different glycoproteins and carbohydrates were tested for inhibition of the interaction (Table 3). Mucin, fetuin and asialofetuin substantially inhibited the adhesion of both MBP-Mub1 and MBP-Mub2 to mucus, whereas the interaction was only slighted affected by the same concentration of BSA. This suggests that Mub interacts with carbohydrates on the mucus components. The inhibition by asialofetuin was of the same order of magnitude as the inhibition by fetuin, which shows that sialic acid is not involved in the interaction. The sugars maltose, glucose, fucose, raffinose, N-acetylgalactosamine, sialic acid, N-acetylglucosamine and mannose were tested for their abilities to inhibit the interaction and most of them did not affect adhesion. The exceptions were fucose that reduced both MBP-Mub1 and MBP-Mub2 adhesion by approximately 10 % and mannose that slightly inhibited MBP-Mub2 adhesion. To investigate the role of Mub in the adhesion of strain 1063 to mucus components, an inhibition experiment with αMub1 and αMub2 was conducted. The antibodies were shown to reduce the binding of strain 1063 to mucus by 69 % from an OD of 0n169p0n023 (p) to %!$ could only be seen if 0n053p0n019, but the inhibition antibodies against both types of repeats of Mub were used together. This strongly indicates that Mub contributes to the adhesion of strain 1063 to mucus in vitro. DISCUSSION

Mub is the first described Lactobacillus adhesion protein that features the same domain organization as most of the characterized adhesins from other Gram-positive bacteria (Fischetti, 2000). The architecture thus includes a signal peptide, C-terminal anchoring sequences, repeated regions with adhesion properties and an Nterminal region that often has an unknown function. However, there are some unusual features. Mub protein is one of the largest bacterial cell-surface proteins identified. Besides that, the corresponding mub gene has two possible translation start sites. A second site is predicted to generate a 35 aa secretion signal peptide with the same predicted cleavage site as the proposed 49 aa signal peptide. The second start is located in the wrong reading frame, with a possible k1 frameshift site located just behind it. Five nucleotides after the second start a TTTC site, known to be highly frameshift-prone (Schwartz & Curran, 1997), is located. Furthermore, the site is followed by a stem–loop structure, which is required for many programmed frameshifts (Farabaugh, 1996). Although speculative, a protein with two possible signal peptides where the second might be regulated by a frameshifting mechanism could represent an interesting mechanism for regulation and localization of secreted proteins. Another uncommon feature is that the membrane anchor sequence at the C terminus of Mub contains a cysteine residue. Rocha & Fischetti (1999) identified and characterized a fibronectin-binding pro439


tein from Streptococcus pyogenes that contained three cysteine residues, one of which was located in the membrane anchor. However, the authors did not postulate any function of the cysteine residues. In Staphylococcus aureus, proteins with the consensus cellwall-anchoring sequence have been shown to be cleaved within the motif LPXTG after transport through the membrane (Schneewind et al., 1995). The presence of a highly functional amino acid such as cysteine in this region could suggest an alternative mechanism for the anchoring of Mub to the cell surface. The fact that strain 1063 adhered efficiently to both pig and hen mucus indicates that the bacterium has little or no host specificity regarding adhesion to mucus. No inhibition of adhesion of strain 1063 cells could be seen when αMub1 or αMub2 antibodies were used separately. However, the significant inhibition observed when using the antibodies together suggests that both types of repeats are important for binding of the bacteria to mucus components. Moreover, since the P108 antibodies, raised against a mix of cell-surface and secreted proteins from strain 1063, blocked binding more efficiently than the mixture of αMub1 and αMub2, other proteins may also be involved in the binding. The presence of two types of amino acid residue repeats and a 500 aa N-terminal region makes it likely that Mub is a multifunctional protein, in common with many described adhesins from Gram-positive organisms (Fischetti, 2000). The two types of repeats both adhere to mucus, mucus fractions and to mucin (Table 2). Western blot experiments showed that the Mub1 and Mub2 repeats adhere to the same components in pig intestinal mucus and to the same components in hen mucus. Furthermore, the inhibition of binding by the glycoproteins fetuin and asialofetuin suggests that Mub adheres to carbohydrate structures in mucus. However, the binding of MBP-Mub1 was less sensitive to the addition of glycoproteins, indicating that it also interacts with other components in mucus. Of the simple carbohydrates tested, only fucose gave a reduction in binding of MBP-Mub1 and MBP-Mub2, indicating that it might form part of a more complex carbohydrate receptor structure. There was also a pH dependence in the binding with maximum binding below neutral pH. Interestingly, MBP-Mub1 and MBP-Mub2 had different pH optima for binding, which could reflect an adaptation to the interaction with different niches in the gastrointestinal tract. Mukai et al. (1998) showed that three Lb. reuteri strains adhered to different sugar residues of glycoconjugates. Specificity of binding was determined for only one of the strains that, in contrast to Mub, exhibited binding that was sensitive to addition of N-acetylgalactosamine. The type 1 and 2 repeats of Mub display extensive sequence similarities with each other and also to the repeats of the Lc. lactis protein YwfG. Lactococci are not usually considered as members of the normal intestinal microflora, but in studies during the 1970s Lc. lactis was commonly isolated from faeces from many different groups of humans (Finegold et al., 1983). The 440

presence of a protein with strong similarities to Mub suggests that Lc. lactis actually is adapted to the gastrointestinal tract environment. The Mub repeats also share similarities with an NANB hepatitis virus antigen (Reyes et al., 1990) and the human ocular component hr44 (Braun et al., 1995). The hepatitis peptide is described as a peptide that is recognized by antiserum from a human infected with the virus and hr44 is a protein located on the cell surface of epithelial cells of the eye. The functions of these proteins are not known, but both might be present at surfaces that are in contact with a mucus layer. The finding that proteins from such distant organisms appear related is intriguing. It could be speculated that the similarities between the proteins reflect some common mechanism for interaction with mucus components. In most cases, the initial step in infection by any microorganism is adhesion or association with host mucosal surfaces. Therefore, blocking of this step is a potential point of disease control. In several reports, probiotic lactobacilli have been shown to inhibit the in vitro attachment of enterovirulent bacteria to mucus and enterocytes and also to prevent invasion of the pathogen (Craven & Williams 1998 ; Tuomola et al., 1999). Whether the mechanism of inhibition is competition for the same receptor structures or an effect of steric hindrance by Lactobacillus cells is not known. Lb. reuteri has been tested for its probiotic activities in different animal systems and has been found to have protecting effects against pathogenic micro-organisms such as Salmonella (Edens et al., 1997) and Cryptosporidium (Alak et al., 1997). Furthermore, children suffering from rotavirus diarrhoea were found to have a significantly reduced duration of the disease when treated with Lb. reuteri (Shornikova et al., 1997). The importance of the mucus-binding properties of lactobacilli in this context has not yet been evaluated. Preliminary results show that mub is not present in a majority of other Lb. reuteri strains tested. However, tested strains of the species Lactobacillus mucosae harboured it (Roos et al., 2000). Construction of a mub mutant is planned in the near future. This is the key for a thorough evaluation of the role of Mub in colonization and in the control of pathogenic micro-organisms. ACKNOWLEDGEMENTS This study was supported by grants from The Swedish National Board for Industrial and Technical Development, NUTEK, (P5974) in a co-operation project with BioGaia AB, Stockholm, Sweden, and from The Swedish Council for Forestry and Agricultural Research, SJFR, (31.0243\91).

REFERENCES Adlerberth, I., Ahrne! , S., Johansson, M.-L., Molin, G., Hanson, L. AH . & Wold, A. E. (1996). A mannose-specific adherence mechan-

ism in Lactobacillus plantarum conferring binding to the human colonic cell line HT-29. Appl Environ Microbiol 62, 2244–2251. Alak, J. I. B., Wolf, B. W., Mdurvwa, E. G., Pimentel-Smith, G. E. & Adeyemo, O. (1997). Effect of Lactobacillus reuteri on intestinal

resistance to Cryptosporidium parvum infection in a murine

Mucus binding of Lactobacillus reuteri model of acquired immunodeficiency syndrome. J Infect Dis 175, 218–221. Aleljung, P. (1994). Collagen binding of intestinal Lactobacillus reuteri : characterisation, purification and cloning. PhD Thesis. Department of Medical Microbiology, University of Lund, Lund, Sweden. Aleljung, P., Shen, W., Rozalska, B., Hellman, U., Ljung, AH . & Wadstro$ m, T. (1994). Purification of collagen binding proteins of

Lactobacillus reuteri NCIB 11951. Curr Microbiol 28, 231–236. Axelsson, L. & Lindgren, S. (1987). Characterisation and DNA homology of Lactobacillus strains isolated from pig intestine. J Appl Bacteriol 62, 433–440. Axelsson, L. T., Chung, T. C., Dobrogosz, W. J. & Lindgren, S. E. (1989). Production of a broad spectrum antimicrobial substance

by Lactobacillus reuteri. Microbial Ecol Health Dis 2, 131–136. Barenkamp, S. J. & St Geme, J. W., III (1996). Identification of a

second family of high-molecular-weight adhesion proteins expressed by non-typable Haemophilus influenzae. Mol Microbiol 19, 1215–1223. Berg, R. D. (1996). The indigenous gastrointestinal microflora. Trends Microbiol 4, 430–435. Bos, E. S., van der Doelen, A. A., van Roy, N. & Schuurs, A. H. (1981). 3,3h,5,5h-Tetramethylbenzidine as an Ames test negative

chromogen for horse-radish peroxidase in enzyme-immunoassay. J Immunoassay 2, 187–204. Brassart, D. & Schiffrin, E. J. (1997). The use of probiotics to reinforce mucosal defence mechanisms. Trends Food Sci Technol 8, 321–326. Braun, G., McKechnie, N. M. & Gu$ rr, W. (1995). Molecular and immunological characterization of hr44, a human ocular component immunologically cross-reactive with antigen Ov39 of Onchocerca volvulus. J Exp Med 182, 1121–1132. Coconnier, M.-H., Klaenhammer, T. R., Kerne' is, S., Bernet, M.-F. & Servin, A. L. (1992). Protein-mediated adhesion of Lactobacillus

acidophilus BG2FO4 on human enterocyte and mucus-secreting cell line in culture. Appl Environ Microbiol 58, 2034–2039. Conway, P. L. & Kjelleberg, S. (1989). Protein-mediated adhesion of Lactobacillus fermentum strain 737 to mouse stomach squamous epithelium. J Gen Microbiol 135, 1175–1186. Craven, S. E. & Williams, D. D. (1998). In vitro attachment of Salmonella typhimurium to chicken cecal mucus : Effect of cations and pretreatment with Lactobacillus spp. isolated from the intestinal tract of chickens. J Food Protein 61, 265–271. Drasar, B. S. & Barrow, P. A. (1985). The bacterial flora of the normal intestine. In Intestinal Microbiology, pp. 19–41. Edited by J. A. Cole, C. J. Knowles & D. Schlessinger. Washington, DC : American Society for Microbiology. Edens, F. W., Parkhurst, C. R., Casas, I. A. & Dobrogosz, W. J. (1997). Principles of ex ovo competitive exclusion and in ovo

administration of Lactobacillus reuteri. Poultry Sci 76, 179–196. Farabaugh, P. J. (1996). Programmed translational frameshifting. Microbiol Rev 60, 103–134. Felsenstein, J. (1993).  (Phylogeny Inference Package) version 3.5c. Seattle : Department of Genetics, University of Washington. Finegold, S. M., Sutter, V. L. & Mathisen, G. E. (1983). Normal indigenous intestinal flora. In Human Intestinal Microflora in Health and Disease, pp. 3–31. Edited by D. J. Hentges. New York : Academic Press. Fischetti, V. A. (2000). Surface proteins of Gram-positive bacteria. In Gram Positive Pathogens, pp. 11–24. Edited by V. A. Fischetti,

R. P. Novick, J. J. Ferretti, D. A. Portnoy & J. I. Rood. Washington, DC : American Society for Microbiology. Henriksson, A. & Conway, P. L. (1992). Adhesion to porcine squamous epithelium of saccharide and protein moieties of Lactobacillus fermentum strain 104-S. J Gen Microbiol 138, 2657–2661. Jonsson, H., Stro$ m, E. & Roos, S. (2001). Addition of mucin to the growth medium triggers mucus-binding activity in different strains of Lactobacillus reuteri in vitro. FEMS Microbiol Lett 204, 19–22. Kirjavainen, P. V., Ouwehand, A. C., Isolauri, E. & Salminen, S. J. (1998). The ability of probiotic bacteria to bind to human

intestinal mucus. FEMS Microbiol Lett 167, 185–189. Kotarski, S. F. & Savage, D. C. (1979). Models for study of the specificity by which indigenous lactobacilli adhere to murine gastric epithelia. Infect Immun 63, 1698–1702. Lindgren, S. E., Swaisgood, H. E., Janolino, V. G., Axelsson, L. T., Richter, C. S., Mackenzie, J. M. & Dobrogosz, W. J. (1992). Binding

of Lactobacillus reuteri to fibronectin immobilised on glass beads. Zentbl Bakteriol 277, 519–528. Molin, G., Jeppsson, B., Johansson, M.-L., Ahrne! , S., Nobaek, S., Sta/ hl, M. & Bengmark, S. (1993). Numerical taxonomy of

Lactobacillus spp. associated with healthy and diseased mucosa of the human intestines. J Appl Bacteriol 74, 314–323. Mukai, T. & Arihara, K. (1994). Presence of intestinal lectinbinding glycoproteins on the cell surface of Lactobacillus acidophilus. Biosci Biotechnol Biochem 58, 1851–1854. Mukai, T., Kaneko, S. & Ohori, H. (1998). Haemagglutinin and glycolipid-binding activities of Lactobacillus reuteri. Lett Appl Microbiol 27, 130–134. Naito, S., Hayashidani, H., Kaneko, K., Ogawa, M. & Benno, Y. (1995). Development of intestinal lactobacilli in normal piglets. J

Appl Bacteriol 79, 230–236. Reyes, G. R., Bradley, D. W., Rabin, L. & Fry, K. E. (1990). Post-

transfusion, non-A, non-B hepatitis virus and antigens. International Patent Application WO 90\00597. Rocha, C. L. & Fischetti, V. A. (1999). Identification and characterization of a novel fibronectin-binding protein on the surface of group A streptococci. Infect Immun 67, 2720–2728. Rojas, M. & Conway, P. L. (1996). Colonization by lactobacilli of piglet small intestinal mucus. J Appl Bacteriol 81, 474–480. Roos, S., Aleljung, P., Robert, N., Lee, B., Wadstro$ m, T., Lindberg, M. & Jonsson, H. (1996). A collagen binding protein from

Lactobacillus reuteri is part of an ABC transporter system? FEMS Microbiol Lett 144, 33–38. Roos, S., Lindgren, S. & Jonsson, H. (1999). Autoaggregation of Lactobacillus reuteri is mediated by a putative DEAD-box helicase. Mol Microbiol 32, 427–436. Roos, S., Karner, F., Axelsson, L. & Jonsson, H. (2000). Lactobacillus mucosae sp. nov., a new species with in vitro mucusbinding activity isolated from pig intestine. Int J Syst Evol Microbiol 50, 251–258. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning : a Laboratory Manual, 2nd edn. Cold Spring Harbor, NY : Cold Spring Harbor Laboratory. Savage, D. C. (1980). Adherence of normal flora to mucosal surfaces. In Bacterial Adherence, pp. 33–59. Edited by E. H. Beachey. London : Chapman and Hall. Schneewind, O., Fowler, A. & Faull, K. F. (1995). Structure of the cell wall anchor of surface proteins in Staphylococcus aureus. Science 268, 103–106. 441


Schwartz, R. & Curran, J. F. (1997). Analyses of frameshifting at

UUU-pyrimidine sites. Nucleic Acids Res 25, 2005–2011. Shornikova, A.-V., Casas, I. A., Mykka$ nen, H., Salo, E. & Vesikari, T. (1997). Bacteriotherapy with Lactobacillus reuteri in rotavirus

effect of probiotic bacteria on the adhesion of pathogens to human intestinal mucus. FEMS Immunol Med Microbiol 26, 137–142.

gastroenteritis. Pediatr Infect Dis J 16, 1103–1107.

Wadstro$ m, T., Anderson, K., Sydow, M., Axelsson, L., Lindgren, S. & Gullmar, B. (1987). Surface properties of lactobacilli isolated

Thompson, J. D., Higgins, D. G. & Gibson, T. J. (1994).   :

from the small intestine of pigs. J Appl Bacteriol 62, 513–520.

improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22, 4673–4680. Tuomola, E. M., Ouwehand, A. C. & Salminen, S. J. (1999). The

442

.................................................................................................................................................

Received 17 April 2001 ; revised 24 September 2001 ; accepted 10 October 2001.